Update app.py

app.py

@@ -1,5 +1,6 @@
 from transformers import MusicgenForConditionalGeneration, AutoProcessor, set_seed
 import torch
+import numpy as np
 import gradio as gr
 
 model = MusicgenForConditionalGeneration.from_pretrained("facebook/musicgen-small")
@@ -9,10 +10,11 @@ device = "cuda:0"
 model.to(device)
 
 sampling_rate = model.audio_encoder.config.sampling_rate
+frame_rate = model.audio_encoder.config.frame_rate
 text_encoder = model.get_text_encoder()
 
 
-def generate_audio(prompt, negative_prompt, guidance_scale=3, seed=0):
+def generate_audio(prompt, negative_prompt, guidance_scale=3, audio_length_in_s=20, seed=0):
     inputs = processor(
         text=[prompt, negative_prompt],
         padding=True,
@@ -22,8 +24,10 @@ def generate_audio(prompt, negative_prompt, guidance_scale=3, seed=0):
     with torch.no_grad():
         encoder_outputs = text_encoder(**inputs)
 
+    max_new_tokens = int(frame_rate * audio_length_in_s)
+
     set_seed(seed)
-    audio_values = model.generate(inputs.input_ids[0][None, :], attention_mask=inputs.attention_mask, encoder_outputs=encoder_outputs, do_sample=True, guidance_scale=guidance_scale, max_new_tokens=1028)
+    audio_values = model.generate(inputs.input_ids[0][None, :], attention_mask=inputs.attention_mask, encoder_outputs=encoder_outputs, do_sample=True, guidance_scale=guidance_scale, max_new_tokens=max_new_tokens)
 
     audio_values = (audio_values.cpu().numpy() * 32767).astype(np.int16)
     return (sampling_rate, audio_values)
@@ -35,6 +39,7 @@ gr.Interface(
         gr.Text(label="Prompt", value="80s pop track with synth and instrumentals"),
         gr.Text(label="Negative prompt", value="drums"),
         gr.Slider(1.5, 10, value=3, step=0.5, label="Guidance scale"),
+        gr.Slider(5, 30, value=15, step=5, label="Audio length in s"),
         gr.Slider(0, 10, value=0, step=1, label="Seed"),
     ],
     outputs=[